Balancing pp driver stage

[dgta]

This came up on one of my amps but is more of a general topology question.

In a typical PP driver stage, like a Williamson, let's say you have a 6SN7 wired as a differential pair, driving the final tubes. The balance of this stage is not assured, even with perfectly matched plate loads. I've observed significant imbalance due to tube halves being unmatched. A trim pot solves the problem quickly, but I'm looking for an "automatic" solution.

Clearly a long(er) tail would help, but let's say a negative supply is not available or practical in an existing amp. I am thinking that putting CCS as plate loads on both tube halves would equalize things nicely, but you would lose any differential action.

Any clever ways of getting auto balance AND keeping the diff action?

[DF96]

CCS in the tail. This is necessary with low-mu valves like 6SN7.

[dgta]

That's the obvious solution but I said no negative supply, so no room for it. Only a few volts available between cathodes and ground.

[tubelab.com]

If you can bias the grids up to 40 volts or so, the cathodes will follow and then you can put a CCS in the tail. You will lose that 40 volts which was across the driver tubes, but this is only important where large drive voltage is required.

[dgta]

Aha! Yes, why didn't I think of that? Sort of moving it toward cathode follower operation.

I assume I would lose not only output swing, but also a good chunk of Av. Probably doable in most cases.

Related question: has anyone here designed/tested simple current sources with low voltage drop? How much performance is lost?

[tubelab.com]

I assume I would lose not only output swing, but also a good chunk of Av.

If you are already running an LTP, improving it with a CCS should not cost much gain. If your tail resistor is small enough so that there is obvious imbalance, yes you will lose some gain.

The LM134/LM334 will work with only 2 or 3 volts but is only good to 40 volts and 10 mA max, so it is only suitable for the input stage. The IXYS ICP10M45 needs about 10 volts PLUS the anticipated maximum signal swing across it. That's where the 40 volt number I quoted came from. I have used it in the first stage with grids at 0 volts and -20 volts to the CCS in the cathodes with excellent results. The plates of the input LTP are directly coupled to the grids of the second stage LTP putting the grids of the second stage at about 80 volts with a CCS to ground in the tail.

[the_manta]

A semiconductor CCS should handle that flawlessly. I'm not an expert in semic. CCS, but on page 134 in M.J. Valve Amplifiers (3.Ed) there are many CCS and two only need 15V across them.
It shouldn't be a problem.

But this hasn't anything to do with cathode follower operation. Since both tubes swing in opposite, there is no feedback voltage created which could cause current feedback. Just usual diff amp operation.

If you want a real automatic solution, go for paraphase. Tubes don't even need to be the same for that

[dgta]

Well, no, there isn't 15V to work with.

A normal differential pair only has on the cathode whatever the grid bias/grid swing requires. In my particular case at hand, the tube is a 6SN7 so the common cathodes are at about 8V. This would vary with tube type but it wouldn't be over 15V.

Also, as tubelab said, you need the CCS voltage PLUS the expected grid swing. Raising the cathode to 40-50V seems like the only viable solution, assuming you have enough B+ available to accomodate the plate swing.

[dgta]

Originally Posted by the_manta

If you want a real automatic solution, go for paraphase. Tubes don't even need to be the same for that

If I'm designing from scratch I don't get myself into this pickle to begin with I'm talking about dealing with an existing amp after the fact.

[smoking-amp]

A center tapped plate balancing inductor would equalize the outputs. Maybe an Edcor XSM.